Java: Pulling in protobuf's faster UTF-8 encoder. (#5035)

* Pulling in protobuf's faster UTF-8 encoder.

* Remove Utf8 unsafe code.

java/com/google/flatbuffers/FlatBufferBuilder.java

@@ -21,11 +21,7 @@ import static com.google.flatbuffers.Constants.*;
 import java.io.IOException;
 import java.io.InputStream;
 import java.nio.*;
-import java.nio.charset.CharacterCodingException;
-import java.nio.charset.CharsetEncoder;
-import java.nio.charset.CoderResult;
 import java.util.Arrays;
-import java.nio.charset.Charset;
 
 /// @file
 /// @addtogroup flatbuffers_java_api
@@ -39,7 +35,6 @@ public class FlatBufferBuilder {
     /// @cond FLATBUFFERS_INTERNAL
     ByteBuffer bb;                  // Where we construct the FlatBuffer.
     int space;                      // Remaining space in the ByteBuffer.
-    static final Charset utf8charset = Charset.forName("UTF-8"); // The UTF-8 character set used by FlatBuffers.
     int minalign = 1;               // Minimum alignment encountered so far.
     int[] vtable = null;            // The vtable for the current table.
     int vtable_in_use = 0;          // The amount of fields we're actually using.
@@ -50,9 +45,8 @@ public class FlatBufferBuilder {
     int num_vtables = 0;            // Number of entries in `vtables` in use.
     int vector_num_elems = 0;       // For the current vector being built.
     boolean force_defaults = false; // False omits default values from the serialized data.
-    CharsetEncoder encoder = utf8charset.newEncoder();
-    ByteBuffer dst;
     ByteBufferFactory bb_factory;   // Factory for allocating the internal buffer
+    final Utf8 utf8;                // UTF-8 encoder to use
     /// @endcond
 
     /**
@@ -62,10 +56,31 @@ public class FlatBufferBuilder {
      * @param bb_factory The factory to be used for allocating the internal buffer
      */
     public FlatBufferBuilder(int initial_size, ByteBufferFactory bb_factory) {
-        if (initial_size <= 0) initial_size = 1;
+        this(initial_size, bb_factory, null, Utf8.getDefault());
+    }
+
+    /**
+     * Start with a buffer of size `initial_size`, then grow as required.
+     *
+     * @param initial_size The initial size of the internal buffer to use.
+     * @param bb_factory The factory to be used for allocating the internal buffer
+     * @param existing_bb The byte buffer to reuse.
+     */
+    public FlatBufferBuilder(int initial_size, ByteBufferFactory bb_factory,
+                             ByteBuffer existing_bb, Utf8 utf8) {
+        if (initial_size <= 0) {
+          initial_size = 1;
+        }
         space = initial_size;
         this.bb_factory = bb_factory;
-        bb = bb_factory.newByteBuffer(initial_size);
+        if (existing_bb != null) {
+          bb = existing_bb;
+          bb.clear();
+        } else {
+          bb = bb_factory.newByteBuffer(initial_size);
+        }
+        bb.order(ByteOrder.LITTLE_ENDIAN);
+        this.utf8 = utf8;
     }
 
    /**
@@ -74,7 +89,7 @@ public class FlatBufferBuilder {
     * @param initial_size The initial size of the internal buffer to use.
     */
     public FlatBufferBuilder(int initial_size) {
-        this(initial_size, new HeapByteBufferFactory());
+        this(initial_size, new HeapByteBufferFactory(), null, Utf8.getDefault());
     }
 
     /**
@@ -94,7 +109,7 @@ public class FlatBufferBuilder {
      *                   the existing buffer needs to grow
      */
     public FlatBufferBuilder(ByteBuffer existing_bb, ByteBufferFactory bb_factory) {
-        init(existing_bb, bb_factory);
+        this(existing_bb.capacity(), bb_factory, existing_bb, Utf8.getDefault());
     }
 
     /**
@@ -105,7 +120,7 @@ public class FlatBufferBuilder {
      * @param existing_bb The byte buffer to reuse.
      */
     public FlatBufferBuilder(ByteBuffer existing_bb) {
-        init(existing_bb, new HeapByteBufferFactory());
+        this(existing_bb, new HeapByteBufferFactory());
     }
 
     /**
@@ -503,27 +518,12 @@ public class FlatBufferBuilder {
     * @return The offset in the buffer where the encoded string starts.
     */
     public int createString(CharSequence s) {
-        int length = s.length();
-        int estimatedDstCapacity = (int) (length * encoder.maxBytesPerChar());
-        if (dst == null || dst.capacity() < estimatedDstCapacity) {
-            dst = ByteBuffer.allocate(Math.max(128, estimatedDstCapacity));
-        }
-
-        dst.clear();
-
-        CharBuffer src = s instanceof CharBuffer ? (CharBuffer) s :
-            CharBuffer.wrap(s);
-        CoderResult result = encoder.encode(src, dst, true);
-        if (result.isError()) {
-            try {
-                result.throwException();
-            } catch (CharacterCodingException x) {
-                throw new Error(x);
-            }
-        }
-
-        dst.flip();
-        return createString(dst);
+        int length = utf8.encodedLength(s);
+        addByte((byte)0);
+        startVector(1, length, 1);
+        bb.position(space -= length);
+        utf8.encodeUtf8(s, bb);
+        return endVector();
     }
 
    /**

java/com/google/flatbuffers/Table.java

@@ -19,11 +19,7 @@ package com.google.flatbuffers;
 import static com.google.flatbuffers.Constants.*;
 import java.nio.ByteBuffer;
 import java.nio.ByteOrder;
-import java.nio.CharBuffer;
-import java.nio.charset.CharacterCodingException;
 import java.nio.charset.Charset;
-import java.nio.charset.CharsetDecoder;
-import java.nio.charset.CoderResult;
 
 /// @cond FLATBUFFERS_INTERNAL
 
@@ -31,23 +27,17 @@ import java.nio.charset.CoderResult;
  * All tables in the generated code derive from this class, and add their own accessors.
  */
 public class Table {
-  private final static ThreadLocal<CharsetDecoder> UTF8_DECODER = new ThreadLocal<CharsetDecoder>() {
-    @Override
-    protected CharsetDecoder initialValue() {
-      return Charset.forName("UTF-8").newDecoder();
-    }
-  };
   public final static ThreadLocal<Charset> UTF8_CHARSET = new ThreadLocal<Charset>() {
     @Override
     protected Charset initialValue() {
       return Charset.forName("UTF-8");
     }
   };
-  private final static ThreadLocal<CharBuffer> CHAR_BUFFER = new ThreadLocal<CharBuffer>();
   /** Used to hold the position of the `bb` buffer. */
   protected int bb_pos;
   /** The underlying ByteBuffer to hold the data of the Table. */
   protected ByteBuffer bb;
+  Utf8 utf8 = Utf8.getDefault();
 
   /**
    * Get the underlying ByteBuffer.
@@ -98,34 +88,10 @@ public class Table {
    * @return Returns a `String` from the data stored inside the FlatBuffer at `offset`.
    */
   protected String __string(int offset) {
-    CharsetDecoder decoder = UTF8_DECODER.get();
-    decoder.reset();
-
     offset += bb.getInt(offset);
     ByteBuffer src = bb.duplicate().order(ByteOrder.LITTLE_ENDIAN);
     int length = src.getInt(offset);
-    src.position(offset + SIZEOF_INT);
-    src.limit(offset + SIZEOF_INT + length);
-
-    int required = (int)((float)length * decoder.maxCharsPerByte());
-    CharBuffer dst = CHAR_BUFFER.get();
-    if (dst == null || dst.capacity() < required) {
-      dst = CharBuffer.allocate(required);
-      CHAR_BUFFER.set(dst);
-    }
-
-    dst.clear();
-
-    try {
-      CoderResult cr = decoder.decode(src, dst, true);
-      if (!cr.isUnderflow()) {
-        cr.throwException();
-      }
-    } catch (CharacterCodingException x) {
-      throw new RuntimeException(x);
-    }
-
-    return dst.flip().toString();
+    return utf8.decodeUtf8(bb, offset + SIZEOF_INT, length);
   }
 
   /**

java/com/google/flatbuffers/Utf8.java

+/*
+ * Copyright 2014 Google Inc. All rights reserved.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.google.flatbuffers;
+
+import java.nio.ByteBuffer;
+
+import static java.lang.Character.MIN_HIGH_SURROGATE;
+import static java.lang.Character.MIN_LOW_SURROGATE;
+import static java.lang.Character.MIN_SUPPLEMENTARY_CODE_POINT;
+
+public abstract class Utf8 {
+
+  /**
+   * Returns the number of bytes in the UTF-8-encoded form of {@code sequence}. For a string,
+   * this method is equivalent to {@code string.getBytes(UTF_8).length}, but is more efficient in
+   * both time and space.
+   *
+   * @throws IllegalArgumentException if {@code sequence} contains ill-formed UTF-16 (unpaired
+   *     surrogates)
+   */
+  public abstract int encodedLength(CharSequence sequence);
+
+  /**
+   * Encodes the given characters to the target {@link ByteBuffer} using UTF-8 encoding.
+   *
+   * <p>Selects an optimal algorithm based on the type of {@link ByteBuffer} (i.e. heap or direct)
+   * and the capabilities of the platform.
+   *
+   * @param in the source string to be encoded
+   * @param out the target buffer to receive the encoded string.
+   */
+  public abstract void encodeUtf8(CharSequence in, ByteBuffer out);
+
+  /**
+   * Decodes the given UTF-8 portion of the {@link ByteBuffer} into a {@link String}.
+   *
+   * @throws IllegalArgumentException if the input is not valid UTF-8.
+   */
+  public abstract String decodeUtf8(ByteBuffer buffer, int offset, int length);
+
+  private static Utf8 DEFAULT;
+
+  /**
+   * Get the default UTF-8 processor.
+   * @return the default processor
+   */
+  public static Utf8 getDefault() {
+    if (DEFAULT == null) {
+      DEFAULT = new Utf8Safe();
+    }
+    return DEFAULT;
+  }
+
+  /**
+   * Set the default instance of the UTF-8 processor.
+   * @param instance the new instance to use
+   */
+  public static void setDefault(Utf8 instance) {
+    DEFAULT = instance;
+  }
+
+  /**
+   * Utility methods for decoding bytes into {@link String}. Callers are responsible for extracting
+   * bytes (possibly using Unsafe methods), and checking remaining bytes. All other UTF-8 validity
+   * checks and codepoint conversion happen in this class.
+   */
+  static class DecodeUtil {
+
+    /**
+     * Returns whether this is a single-byte codepoint (i.e., ASCII) with the form '0XXXXXXX'.
+     */
+    static boolean isOneByte(byte b) {
+      return b >= 0;
+    }
+
+    /**
+     * Returns whether this is a two-byte codepoint with the form '10XXXXXX'.
+     */
+    static boolean isTwoBytes(byte b) {
+      return b < (byte) 0xE0;
+    }
+
+    /**
+     * Returns whether this is a three-byte codepoint with the form '110XXXXX'.
+     */
+    static boolean isThreeBytes(byte b) {
+      return b < (byte) 0xF0;
+    }
+
+    static void handleOneByte(byte byte1, char[] resultArr, int resultPos) {
+      resultArr[resultPos] = (char) byte1;
+    }
+
+    static void handleTwoBytes(
+        byte byte1, byte byte2, char[] resultArr, int resultPos)
+        throws IllegalArgumentException {
+      // Simultaneously checks for illegal trailing-byte in leading position (<= '11000000') and
+      // overlong 2-byte, '11000001'.
+      if (byte1 < (byte) 0xC2
+              || isNotTrailingByte(byte2)) {
+        throw new IllegalArgumentException("Invalid UTF-8");
+      }
+      resultArr[resultPos] = (char) (((byte1 & 0x1F) << 6) | trailingByteValue(byte2));
+    }
+
+    static void handleThreeBytes(
+        byte byte1, byte byte2, byte byte3, char[] resultArr, int resultPos)
+        throws IllegalArgumentException {
+      if (isNotTrailingByte(byte2)
+              // overlong? 5 most significant bits must not all be zero
+              || (byte1 == (byte) 0xE0 && byte2 < (byte) 0xA0)
+              // check for illegal surrogate codepoints
+              || (byte1 == (byte) 0xED && byte2 >= (byte) 0xA0)
+              || isNotTrailingByte(byte3)) {
+        throw new IllegalArgumentException("Invalid UTF-8");
+      }
+      resultArr[resultPos] = (char)
+                                 (((byte1 & 0x0F) << 12) | (trailingByteValue(byte2) << 6) | trailingByteValue(byte3));
+    }
+
+    static void handleFourBytes(
+        byte byte1, byte byte2, byte byte3, byte byte4, char[] resultArr, int resultPos)
+        throws IllegalArgumentException{
+      if (isNotTrailingByte(byte2)
+              // Check that 1 <= plane <= 16.  Tricky optimized form of:
+              //   valid 4-byte leading byte?
+              // if (byte1 > (byte) 0xF4 ||
+              //   overlong? 4 most significant bits must not all be zero
+              //     byte1 == (byte) 0xF0 && byte2 < (byte) 0x90 ||
+              //   codepoint larger than the highest code point (U+10FFFF)?
+              //     byte1 == (byte) 0xF4 && byte2 > (byte) 0x8F)
+              || (((byte1 << 28) + (byte2 - (byte) 0x90)) >> 30) != 0
+              || isNotTrailingByte(byte3)
+              || isNotTrailingByte(byte4)) {
+        throw new IllegalArgumentException("Invalid UTF-8");
+      }
+      int codepoint = ((byte1 & 0x07) << 18)
+                          | (trailingByteValue(byte2) << 12)
+                          | (trailingByteValue(byte3) << 6)
+                          | trailingByteValue(byte4);
+      resultArr[resultPos] = DecodeUtil.highSurrogate(codepoint);
+      resultArr[resultPos + 1] = DecodeUtil.lowSurrogate(codepoint);
+    }
+
+    /**
+     * Returns whether the byte is not a valid continuation of the form '10XXXXXX'.
+     */
+    private static boolean isNotTrailingByte(byte b) {
+      return b > (byte) 0xBF;
+    }
+
+    /**
+     * Returns the actual value of the trailing byte (removes the prefix '10') for composition.
+     */
+    private static int trailingByteValue(byte b) {
+      return b & 0x3F;
+    }
+
+    private static char highSurrogate(int codePoint) {
+      return (char) ((MIN_HIGH_SURROGATE - (MIN_SUPPLEMENTARY_CODE_POINT >>> 10))
+                         + (codePoint >>> 10));
+    }
+
+    private static char lowSurrogate(int codePoint) {
+      return (char) (MIN_LOW_SURROGATE + (codePoint & 0x3ff));
+    }
+  }
+
+  // These UTF-8 handling methods are copied from Guava's Utf8Unsafe class with a modification to throw
+  // a protocol buffer local exception. This exception is then caught in CodedOutputStream so it can
+  // fallback to more lenient behavior.
+  static class UnpairedSurrogateException extends IllegalArgumentException {
+    UnpairedSurrogateException(int index, int length) {
+      super("Unpaired surrogate at index " + index + " of " + length);
+    }
+  }
+}

java/com/google/flatbuffers/Utf8Old.java

+/*
+ * Copyright 2014 Google Inc. All rights reserved.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.google.flatbuffers;
+
+import java.nio.ByteBuffer;
+import java.nio.CharBuffer;
+import java.nio.charset.CharacterCodingException;
+import java.nio.charset.CharsetDecoder;
+import java.nio.charset.CharsetEncoder;
+import java.nio.charset.CoderResult;
+import java.nio.charset.StandardCharsets;
+
+/**
+ * This class implements the Utf8 API using the Java Utf8 encoder. Use
+ * Utf8.setDefault(new Utf8Old()); to use it.
+ */
+public class Utf8Old extends Utf8 {
+
+  private static class Cache {
+    final CharsetEncoder encoder;
+    final CharsetDecoder decoder;
+    CharSequence lastInput = null;
+    ByteBuffer lastOutput = null;
+
+    Cache() {
+      encoder = StandardCharsets.UTF_8.newEncoder();
+      decoder = StandardCharsets.UTF_8.newDecoder();
+    }
+  }
+
+  private static final ThreadLocal<Cache> CACHE =
+      ThreadLocal.withInitial(() -> new Cache());
+
+  // Play some games so that the old encoder doesn't pay twice for computing
+  // the length of the encoded string.
+
+  @Override
+  public int encodedLength(CharSequence in) {
+    final Cache cache = CACHE.get();
+    int estimated = (int) (in.length() * cache.encoder.maxBytesPerChar());
+    if (cache.lastOutput == null || cache.lastOutput.capacity() < estimated) {
+      cache.lastOutput = ByteBuffer.allocate(Math.max(128, estimated));
+    }
+    cache.lastOutput.clear();
+    cache.lastInput = in;
+    CharBuffer wrap = (in instanceof CharBuffer) ?
+                          (CharBuffer) in : CharBuffer.wrap(in);
+    CoderResult result = cache.encoder.encode(wrap, cache.lastOutput, true);
+    if (result.isError()) {
+      try {
+        result.throwException();
+      } catch (CharacterCodingException e) {
+        throw new IllegalArgumentException("bad character encoding", e);
+      }
+    }
+    return cache.lastOutput.remaining();
+  }
+
+  @Override
+  public void encodeUtf8(CharSequence in, ByteBuffer out) {
+    final Cache cache = CACHE.get();
+    if (cache.lastInput != in) {
+      // Update the lastOutput to match our input, although flatbuffer should
+      // never take this branch.
+      encodedLength(in);
+    }
+    out.put(cache.lastOutput);
+  }
+
+  @Override
+  public String decodeUtf8(ByteBuffer buffer, int offset, int length) {
+    CharsetDecoder decoder = CACHE.get().decoder;
+    decoder.reset();
+    buffer = buffer.duplicate();
+    buffer.position(offset);
+    buffer.limit(offset + length);
+    try {
+      CharBuffer result = decoder.decode(buffer);
+      result.flip();
+      return result.toString();
+    } catch (CharacterCodingException e) {
+      throw new IllegalArgumentException("Bad encoding", e);
+    }
+  }
+}

pom.xml

@@ -5,7 +5,7 @@
   <modelVersion>4.0.0</modelVersion>
   <groupId>com.google.flatbuffers</groupId>
   <artifactId>flatbuffers-java</artifactId>
-  <version>1.10.0</version>
+  <version>1.10.1-SNAPSHOT</version>
   <packaging>bundle</packaging>
   <name>FlatBuffers Java API</name>
   <description>
@@ -78,6 +78,10 @@
         <groupId>org.apache.maven.plugins</groupId>
         <artifactId>maven-javadoc-plugin</artifactId>
         <version>2.9.1</version>
+        <configuration>
+          <additionalparam>-Xdoclint:none</additionalparam>
+          <additionalOptions>-Xdoclint:none</additionalOptions>
+        </configuration>
         <executions>
           <execution>
             <id>attach-javadocs</id>